EnhanceNet TPS calculator

Open weights Max Planck Institute for Intelligent Systems 814.5K parameters December 2016

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 45,262 tok/s

Fastest card

B200

4,160,080 tok/s · 180 GB

Which GPUs can run EnhanceNet?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
4,160,080 tok/s 2,496,048–6,656,128 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
4,160,080 tok/s 2,496,048–6,656,128 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
3,321,928 tok/s 1,993,157–5,315,084 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
3,321,928 tok/s 1,993,157–5,315,084 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
2,656,731 tok/s 1,594,039–4,250,770 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
2,542,849 tok/s 1,525,709–4,068,558 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
2,542,849 tok/s 1,525,709–4,068,558 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
2,433,647 tok/s 1,460,188–3,893,835 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
2,159,861 tok/s 1,295,917–3,455,778 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
2,159,861 tok/s 1,295,917–3,455,778 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
2,159,861 tok/s 1,295,917–3,455,778 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
2,048,839 tok/s 1,229,304–3,278,143 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
1,747,234 tok/s 1,048,340–2,795,574 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
1,747,234 tok/s 1,048,340–2,795,574 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
1,747,234 tok/s 1,048,340–2,795,574 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
1,747,234 tok/s 1,048,340–2,795,574 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
1,747,234 tok/s 1,048,340–2,795,574 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
1,330,394 tok/s 798,236–2,128,630 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
1,330,394 tok/s 798,236–2,128,630 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
1,108,661 tok/s 665,197–1,773,858 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
1,085,001 tok/s 651,001–1,736,001 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
1,060,820 tok/s 636,492–1,697,313 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
1,060,820 tok/s 636,492–1,697,313 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
1,060,820 tok/s 636,492–1,697,313 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
1,060,820 tok/s 636,492–1,697,313 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: Max Planck Institute for Intelligent Systems
Organisation type: Academia
Country: Germany
Published: 23 December 2016
Authors: Mehdi S. M. Sajjadi, B. Scholkopf, M. Hirsch

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Vision
Task: Image super-resolution
Approach: Supervised

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 814.5K
Training data: 9,830,400,000 tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 1.3 × 10¹⁷ FLOP
How it was established: Hardware

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Training hardware: NVIDIA Tesla K40c
Chips used: 1
Wall-clock time: 24 hours
Power draw: 282 W

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)
Training code: Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Why it is tracked: SOTA improvement,Highly cited
Record confidence: Confident

Sources

Where this record came from and when it was last checked.

Reference: EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run EnhanceNet

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run EnhanceNet

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

The hardware side

Minimum card

Tesla C1080

Memory needed

0.7 GB

Fastest

4,160,080 tok/s

EnhanceNet is small enough at 814.5K parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

At the low end, a Tesla C1080 handles it — 4 GB, at Q8_0, for about 45,262 tokens per second.

At the other end, a B200 generates roughly 4,160,080 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

Background

EnhanceNet was published by Max Planck Institute for Intelligent Systems, in Germany, in December 2016. academia is the category the publisher falls under.

It works in Vision, and is recorded as doing image super-resolution.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Reading the throughput figures

Half the cards that hold it manage more than 116,815.0 tokens per second, and 818 exceed reading speed outright.

Every weight participates in every token here, so bandwidth is the whole story: the ranking below is effectively a ranking of memory throughput.

Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

How it was trained

Training it took roughly 1.3 × 10¹⁷ FLOP of computation, on NVIDIA Tesla K40c — a measure of what producing the model cost, not of how fast it answers.

It was trained on about 9,830,400,000 tokens of text.

It is tracked in the underlying dataset for one reason in particular: sOTA improvement,Highly cited.

Step by step

How to choose a GPU for EnhanceNet

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Read the memory figure first

Every card here has been checked against EnhanceNet — around 0.7 GB at Q8_0. Capacity is the gate — a card either holds it or it does not.
02

Decide how long your conversations run

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason EnhanceNet stops fitting a card that seemed fine.
03

Set a quality floor

Compression is what makes EnhanceNet fit smaller cards, at some cost in accuracy — Q8_0 on the smallest card that fits. A minimum quality removes the ones that go too far.
04

Compare tokens per second, not specifications

Ranking by tokens per second for EnhanceNet follows memory bandwidth, not core counts, which is why the B200 tops it at 4,160,080 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage EnhanceNet from those with room to spare. Buy for the second if the context might grow.
06

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once EnhanceNet is settled.

Answers

EnhanceNet — common questions

How accurate are these EnhanceNet speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 2,496,048–6,656,128 tok/s on the B200 rather than a single number.

What GPU do I need to run EnhanceNet?

The smallest card in our catalogue that holds EnhanceNet is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 45,262 tokens per second. 818 cards in total can run it.

How fast is EnhanceNet on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 4,160,080 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run EnhanceNet clear that.

How much VRAM does EnhanceNet need?

About 0.7 GB at Q8_0 compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Can I run EnhanceNet on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.7 GB and generating roughly 774,815 tokens per second — a comfortable fit.

Can I run EnhanceNet on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.7 GB and generating roughly 474,457 tokens per second — a comfortable fit.

Can I run EnhanceNet on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.7 GB and generating roughly 587,611 tokens per second — a comfortable fit.

Can I run EnhanceNet on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.7 GB and generating roughly 696,813 tokens per second — a comfortable fit.

Is EnhanceNet open source?

Its weights are published, so EnhanceNet can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does EnhanceNet have?

EnhanceNet has 814.5K parameters. 2*3*3*3*64+22*3*3*64*64=814464 24 CNN layers with 3x3 kernels and 64 channels and 3 input/output channels (see Table 1). That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created EnhanceNet?

EnhanceNet was published by Max Planck Institute for Intelligent Systems, based in Germany, categorised as academia.

When was EnhanceNet released?

EnhanceNet was published in December 2016. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is EnhanceNet used for?

EnhanceNet works in Vision, and is recorded as handling image super-resolution. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

Where can I download EnhanceNet?

The weights for EnhanceNet are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

How much compute was used to train EnhanceNet?

Around 1.3 × 10¹⁷ FLOP, on NVIDIA Tesla K40c. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Can I run EnhanceNet if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part in system memory drags the whole thing down. Our figures for EnhanceNet assume it is fully resident.

Would two GPUs run EnhanceNet faster?

Two cards buy memory rather than speed. That matters for EnhanceNet only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for EnhanceNet?

Because capacity varies, so does how hard EnhanceNet has to be squeezed — 1 distinct levels appear in the table above. Set a minimum quality to compare at one.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models